Amorim Lopes et al. Human Resources for Health (2015) 13:38 DOI 10.1186/s12960-015-0028-0

REVIEW Open Access Handling healthcare workforce planning with care: where do we stand? Mário Amorim Lopes1*, Álvaro Santos Almeida2 and Bernardo Almada-Lobo1

Abstract Background: Planning the health-care workforce required to meet the health needs of the population, while providing service levels that maximize the outcome and minimize the financial costs, is a complex task. The problem can be described as assessing the right number of people with the right skills in the right place at the right time, to provide the right services to the right people. The literature available on the subject is vast but sparse, with no consensus established on a definite methodology and technique, making it difficult for the analyst or policy maker to adopt the recent developments or for the academic researcher to improve such a critical field. Methods: We revisited more than 60 years of documented research to better understand the chronological and historical evolution of the area and the methodologies that have stood the test of time. The literature review was conducted in electronic publication databases and focuses on conceptual methodologies rather than techniques. Results: Four different and widely used approaches were found within the scope of supply and three within demand. We elaborated a map systematizing advantages, limitations and assumptions. Moreover, we provide a list of the data requirements necessary to implement each of the methodologies. We have also identified past and current trends in the field and elaborated a proposal on how to integrate the different methodologies. Conclusion: Methodologies abound, but there is still no definite approach to address HHR planning. Recent literature suggests that an integrated approach is the way to solve such a complex problem, as it combines elements both from supply and demand, and more effort should be put in improving that proposal. Keywords: Review, Health-care workforce planning, Supply, Demand, Needs, Health policy

Introduction services and the visits are shorter [5]; work overload of Health-care human resources (HHR) planning has been the available physicians and nurses, resulting in sleep- identified as the most critical constraint in achieving the deprivation, ultimately compromising patient safety [6]; well-being targets set forth in the United Nations’ Mil- and queues and waiting lists resulting from insufficient lennium Development Goals [1]. Moreover, the effective medical staff, causing avoidable patient deaths [7]. use and deployment of personnel is paramount to ensure Another argument supporting HHR planning is the an efficient service delivery in terms of cost, quality and recent rise in health-care expenditure, both in per capita quantity [2]. Failure to do so may result in an oversupply spending on health and as a proportion of per capita or shortage of clinical staff. While the former may lead to domestic product in real terms [8]. The average annual economic inefficiencies and misallocated resources under growth rate of health-care expenditure in a selection of the guise of unemployment [3] or inflated costs through 18 countries that are part of the Organisation for Eco- supplier-induced demand [4], the latter is linked to a more nomic Co-operation and Development (OECD) was 3.0 % extensive list of negative effects, including but not limited between 1980 and 1990 and 3.3 % in the decade after to the following: lower quantity and quality of medi- [8]. Recent studies confirm the rising trend, with health cal care as few resources exist to provide the necessary spending growing at an average of 3.8 % in 2008 and 3.5 % in 2009 [9], well above the growth rate of the *Correspondence: [email protected] gross domestic product. Health worker wages account for 1INESC TEC, Faculdade de Engenharia, Universidade do Porto, Porto, Portugal Full list of author information is available at the end of the article

© 2015 Amorim Lopes et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http:// creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 2 of 19

about 50 % of total public and private health expenditure Literature search method across several countries [5], meaning that cost contain- We carried out an extensive literature review, includ- ment and efficiency improvements will necessarily require ing academic research papers and technical reports from the involvement of the workforce. institutions such as the OECD or WHO. Selected papers In sharp contrast to other scientific areas where a set date between 1951 and 2013, and the results were of well-defined methodologies and techniques is gener- reported in a chronological and evolutionary way so as ally adopted and refined to solve a given problem, in to clearly identify methodologies that are still in use to HHR planning, methodologies (the conceptual scope of this day. The search methodology can be summarized as analysis) and approaches (the techniques applied upon a follows: after selecting a set of search terms and gener- particular method) abound, and there is still no commonly ating reliable combinations, we used electronic research accepted or favoured procedure to accurately forecast databases to search for related articles. We then selected physician requirements [3, 10]. The methodologies fol- a maximum of 20 papers for each combination of search lowed by countries vary significantly, in some cases with terms, including the 10 most cited, the 5 most recent and no long-term strategic HHR planning at all, but a wide 5 that were randomly chosen. A backward/forward search array of options does not seem to be a determining factor was conducted, and the abstract was analysed to ensure in improving the accuracy of forecasting [11]. Despite the that the papers met the search criteria. Papers that failed lack of focus, the accuracy of the projections appears to be to meet any of the search criteria were excluded. making progress in some cases, as a review reporting the To identify search terms, we consulted the available case of The Netherlands shows [12], an encouraging sign literature reviews and technical reports [5, 10, 11, 13] to the ongoing research. so as to a obtain a list of key terms frequently used in A definite approach to the problem, or at least a sta- this research field. Table 1 displays the search terms more ble starting block, will require a comprehensive overview frequently employed in the literature. Multiple combi- of how the problem has been tackled since its inception. nations were selected using these key search terms. For For this purpose, we provide a thorough analysis of the instance, all possible combinations of health and health- field, to lay down the foundations for future research, cou- care with (AND) workforce, manpower, physicians, nurses pled with a historical perspective on the development of and (AND) forecast, projection, planning.Relatedsubor- the HHR literature, analysing how the field has evolved dinate queries such as physicians supply forecast, nurses and what methodologies have emerged and continue to be supply forecast, healthcare supply forecast, healthcare employed. Secondly, we analyse the strengths and pitfalls demand forecast were also employed. These terms were of each of the methodologies and provide a data require- then used on the online databases PubMed, MEDLINE, ment framework containing all the variables and data that Embase, ProQuest, Healthstar, ABI/Inform, INSPEC, need to be taken into account in order to address the Google Scholar and Scopus to obtain a base set of the 10 problem thoroughly. The review is selective as it focuses most cited, 5 most recent and 5 randomly chosen papers. primarily on articles that seem to have had a clear impact Of this initial selection, an abstract matching and back- on the evolution of the field, although broad in scope ward/forward search was conducted to assess whether the as it attempts to extensively describe all known meth- topic covered was relevant. Publications that failed to ver- ods. Finally, it describes where we stand and the road ify these criteria were excluded. A total of 308 publications ahead, providing a brief overview of new and emerging were retrieved, with 75 meeting at least 1 of the inclusion approaches to the HHR planning problem. criteria using the combination of search terms and were To the best of our knowledge, the last comprehensive thus included in this review. Table 2 describes our search academic paper on the subject dates back to 1978 [13]. methodology. Literature reviews exist but tend to either focus on a par- ticularperiodoronasubsetofthemethodologiesor Scope techniques [11, 14] or to be framed as technical reports HHR planning is a comprehensive field far extending aimed at a wider readership, such as the OECD’s extensive the number of physicians and nurses. Other health-care review of 26 projection models used in 18 countries [9] workers such as hygienists, therapists, managers, admin- or WHO’s policy recommendations to the EU [15]. The istrative assistants and other support staff also play a literature reviews can also consist of a technical report tar- critical role, relieving the clinical staff of bureaucratic geting a country in particular [16]. In fact, some authors and time-consuming tasks. In fact, skill-mix studies show point out that more systematic reviews, assessments of that proper task delegation is critical to ensure proper potential interventions and further research to aid policy health-care delivery. Furthermore, a complete assessment makers are highly needed [17]. This paper aims to narrow may also require the analysis of the impact of other indi- this gap by being a starting point both for academics and rect stakeholders, such as workforce educators, regula- policy makers. tors, funders and employers. Assessing how the training Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 3 of 19

Table 1 Key terms used to conduct the search existing hospitals and health-care units absorb a planned Keywords Search queries increase in the number of health-care professionals?) is a Health Workforce planning critical requirement for a well-guided policy. Without disregarding the importance of these other Healthcare Healthcare forecasting professions, in this paper, we will focus solely on review- Workforce Health human resources ing the planning of the clinical staff that directly pro- Manpower Health manpower vide health-care services and, more specifically, on the Physicians Health planning physicians and nurses, along with references to related Nurses Healthcare planning fields like dentistry. Obtaining reliable projections for the Forecast Health services available and necessary human resources is an obligatory starting point. Moreover, the prominence will be in the Projection Health supply spectrum of different methodologies that may be used to Planning Health demand obtain forecasts for the number of physicians and nurses, ... Healthcare needs with short references to the approaches or technical appa- Healthcare providers ratus, commonly used to apply a given methodology a. Physician forecasting Also, our concern is HHR planning only at the national Nurse forecasting and regional level. HHR planning at a local level (hospi- tal or medical centre) is conceptually different, involving Nursing staff other methodologies and tools, and therefore, it is not Manpower inserted in this paper. Manpower planning The remainder of this paper is organized as follows: Workforce forecasting in the “Background” section, we introduce the general Workforce projections and governing principles that characterize the health-care Workforce management market. The background information provided is critical to equip the reader with the necessary concepts. In the Staff levels “Evolution of the field” section, we proceed with an evolu- Health staffing levels tionary and chronological description of the field, expos- Shortage healthcare workers ing the work and methodologies that have been shaping the research field. In the “Discussion” section, we discuss the current trends in this research area and the road ahead is conducted (i.e. could the training time be reduced?; do regarding future research directions. We also present a medical schools have the capacity to train a given num- summary of all the findings, including a table with an ber of trainees?; are more medical schools necessary?), overview of the methodologies and a data-requirement the impact of regulatory requirements (i.e. is the entry framework to understand which methodologies can be to medical school limited by government-fixed numerus used based on the data available, as well as a proposal sug- clausus?) or financial and service constraints (i.e. can the gesting a way to develop an integrated approach. Finally, we finish with a brief summary and conclusion. Table 2 The search method applied in this review Background Step Search method HHR planning as a scientific area and topic of theo- 1 Identify common search terms from reviews, books and retical and applied research evolved significantly from technical papers non-existence into a remarkable and serious effort of pri- 2 Generate plausible combinations of terms to be used for vate and governmental institutions, which tried to antic- search using the key search terms identified ipate how many human resources, primarily physicians 4 Search for these terms on PubMed, MEDLINE, Embase, and nurses, will be necessary in order to maintain or ProQuest, Healthstar, ABI/Inform, INSPEC, Google Scholar and Scopus even improve the quantity, quality, availability and effec- tiveness of the medical services provided. Improved life 5 Select a base set for the results consisting of the 20 papers (10 most cited, 5 most recent and 5 randomly chosen) expectancy and changing demographics, epidemiological 6 Match the abstract and perform a forward and backward trends, improved socio-economic conditions and an ever- search to verify the relevance of the paper for the selected base increasing world population may result in a rise in the set expected demand for health-care services [18] and, there- 7 Exclude papers that address none of the topics covered, that fore, further additions to the list of patients of an ageing only make a brief reference to the subject at hand or that are medical workforce [19]. It then comes as no surprise that not written in English health workers are recognized as a critical resource for Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 4 of 19

achieving population health goals [1], working at the front Despite the limitations, some measures to overcome gate of the health-care sector. imbalances in the quantity (number) of physicians and The health-care sector is an intricate, albeit funda- nurses have already been identified in the health policy mental, part of ancient and modern societies, and it literature [17, 21], namely the following: increasing the comprises a long list of agents, from the individual seek- number of domestic- and foreign-trained medical grad- ing health-care services to the medical staff providing uates or increasing the number of medical schools and them, all operating within a legal framework involving classroom sizes; increasing the enrolment limits (numerus providers, consumers, insurance companies, government, clausus); reducing the requirements for entry to medical medical schools and regulatory institutions. Regardless schools; raising the wages of the medical staff, as well as of the statutory system in place, either a Bismarckian- the perspectives for their future career path; or reducing based or a Beveridgean-based organization, at its core, the costs of attending medical school, which may encour- the health-care market is always composed of both sup- age potential students to enrol. In Table 3, we provide a pliers of health services and patients demanding their more extensive list of policies to cope with a shortage in services. On the one side is the workforce of physi- the number of health workers. These proposals are short- cians, nurses and remaining clinical staff trained and term measures to alleviate the immediate stress put on ready to assist those in need. On the other side stand the health-care system triggered by an undersupply of per- the forces that drive the demand for medical services, sonnel and may not be suitable for tackling long-term strongly related to demographic, socioeconomic and epi- imbalances due to huge shortages or surpluses of medical demiological factors. Analysing these two market forces staff. is a critical step in assessing whether the available health- Still within the scope of supply, other approaches for care human resources are enough in quantity and skills handling the problem of insufficient human resources to meet the current and future demand in due time and have also been suggested, addressing the problem from may lay solid foundations for further research, considering perhaps changes to the existing health policy framework. Despite the similarities, the health-care market diverges Table 3 Health policy options for targeting health workforce from a traditional market of goods and services for sev- imbalances and alter health-care outcomes (adapted from [17] eral reasons [20]. A high degree and extent of uncertainty and [86]) affects both supply and demand; asymmetric information Field Policy option between physicians and patients, restrictions on competi- tion, strong government interference and supply-induced Education Increase numbers of new students demand are some of the most glaring differences that can Recruit foreign graduates be pinpointed. These may be relevant when assessing the Recognize previous learning impact of any policy involving HHR planning. Improve curriculum content Regulatory Recognize overseas qualifications Supply Introduce temporary employment Supplying human capital with the appropriate expertise so regulations as to enable workers to perform and satisfy the demand for Subsidized education for return of health care is no simple task. The time and effort required service to equip HHR, especially physicians and advanced nurse Enhanced scope of practice practitioners, exceeds that of most other professions. In some particular health-care professions, the set of nec- Different types of health workers essary skills to qualify for medical practice is acquired Financial incentives Increase trainee salaries through extensive academic learning which involves the Raise wages enrolment in long courses that may take up decades to Provide non-wage benefits complete due to a strict licencing process. Introduce incentives for return of A considerable amount of HHR studies focus solely on skilled migrants this approach, basing their research on the estimation of Establish retirement policies the expected supply of physicians by accounting for the Employ lay health workers intakes, exits, migrations and population growth in order Professional and personal support Better living conditions to maintain the present ratio of practitioners, using “stock- and-flow” models for that purpose [3]. The analysis of the Safe and supportive working environment medical training process is relevant but may be insuffi- cient, as several other factors may affect the efficiency and Career development programmes effectiveness of the care services delivered. Public recognition measures Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 5 of 19

an angle besides medical training. For instance, the com- long as data is available, the method lacks the fine detail position of the core competences and activities of the of such a complex system, ignoring other factors such as physicians, the skill mix, may be reorganized to enhance needs, demand or institutional frameworks that may have the roles performed by the clinical staff, relieving them an influence on the productivity of countries or regions from tasks that could be safely assigned to other health- with similar worker-to-population ratios. Moreover, it care professionals [22]. This strategy does not require a abstains from exposing the causes for such asymmetries or changeinthenumberofphysiciansbuttherestructuring from evaluating the efficiency of the available workforce. of the available human resources and medical compe- tences. Complementarily, supporting policies and reforms Demand that enhance the productivity, that is, the ratio of out- Demand for health care is a derived demand [25], which put per unit of input given a certain level of technology means that people do not seek health care services as a and methodology, of the medical staff may result in an final good for consumption but as an intermediate service increased outcome that also does not require a change allowing them to be healthy and to improve their stock in the quantity of labour workforce [23]. Assessing the of health capital (well-being). They want to improve their productivity of the clinical staff is now quite common health, and to do so, they seek health-care services. As [24], and operations research applied to the improvement in other markets, the determinants of aggregate demand of patient flows, queueing, master surgery scheduling, for health-care services are population size, income and ambulance fleet management and staff rostering may play preferences. Moreover, for countries where medical care is a very important role in increasing current levels of pro- mostly an out-of-pocket expenditure, demand is restricted ductivity. In summary, the initial focus of supply-based by the patients’ ability to pay. If a patient requires medical methodologies was on the training process. As of late, attention and is unable to finance it, this need for health more focus has been given to the productivity and to the care will not translate into effective demand, despite its skill mix of the labour workforce as well. existence. Accounting for these cases is especially impor- tant in countries where health care is not publicly subsi- Methodologies for modelling supply dized or where there are obstacles to entry other than the Training (entries and losses) The purpose is to model the availability of resources. training process so as to predict the number of entrants The concept of needs in health care is not consensual in each year. This way, and in combination with migra- in the health literature, with a semantic confusion arising tory flows, mortality, exit and drop out rates, it becomes from its use in health economics [13, 26]. While the eco- possible to estimate the number of physicians and nurses nomic or effective demand translates the actual, observed available for each year, with everything else held constant. demand, usually measured in terms of service utilization Productivity The productivity of the medical workforce is ratios (such as bed occupancy rates, number of inpa- not constant, as some professionals work harder or better tients), the needs component tries to fully encompass the than others or simply because there is an excess of bureau- epidemiological conditions that characterize a given pop- cracy to comply with. Without touching on the quantity ulation, measured through morbidity and mortality rates of professionals, it is possible to reorganize services and or by the opinion of a panel of experts, and how that incentives so as to promote increased productivity or may translate into a given quantity of required health-care implement lean and operations research recommenda- services. Therefore, we see that the classical concept of tions to significantly improve the output and outcome of economic demand may not reflect the biological needs of the workforce. the population, as it may leave out the necessities of the Skill mix Since a degree of interdisciplinarity exists population regardless of their ability to pay. In the needs between medical professionals, it is possible to reassess component, the emphasis is on the medical conditions the tasks performed by each professional, relieving physi- that may lead to demand for health care, deriving from the cians from day-to-day bureaucratic routines or review- evolution of chronic diseases, prevalence rates and over- ing the competences of the nursing profession so as to all morbidity patterns. This distinction is better illustrated broaden their scope of action. Horizontal substitution in Fig. 1, where we present the case when all demand is (between different medical specialties) and vertical substi- met, at a given price, and equilibrium is attained. The- tution (between different working classes) can be used to oretical demand, projected strictly in terms of biological improve the amount of health-care services provided. needs without a budget constraint (either households’ Worker-to-population ratios This method establishes a income or public budget), may not always correspond to desired ratio for the number of physicians and nurses per the demand effectively observed. The reason being that unit of population and compares it to the actual ratios. the quantity sought is limited by the disposable income Policies to increase or decrease these ratios may then be directed towards out-of-pocket health expenditure or by pushed forward. Although simple and easy to apply as limits to the government budget that is allocated to health Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 6 of 19

is expected that more endocrinologists will be necessary to assist with the treatments. The raw definition of needs is not subject to any boundaries other than those set by epidemiological constraints and medical advances. A substantial part of the studies targeting supply hold current demand constant, thereby leaving out a proper analysis of what drives demand for health care. In fact, a change in the factors that influence demand or the emergence of new health conditions in a population may require a reorganization in the quantity, composition and skill mix of the medical workforce to ensure that all sup- ply meets demand. This suggests that targeting the right number of people and the right skills depends as much on the health conditions and epidemiological characteristics of a given population as on the supply of physicians and nurses [33]. In summary, three methods are commonly used to anal- yse HHR planning from a demand-based perspective [13]. Fig. 1 Law of supply and demand applied to health services. The Most of the methods build upon the definitions of needs health-care market depicted in terms of supply and demand, with a and effective demand, and some overlap in their scope of tentative distinction between potential and effective demand application. Contrarily to the approaches found in supply- based methodologies, where the object of study remains thesameandalternativeanalyticalmethodsareemployed, care. We draw the distinction by plotting both the curve of in demand, opting for a different method may change the needs (potential demand), corresponding to a no gap sce- scope of the analysis. nario, and the economic (effective) demand that is actually observed. Although needs is a fundamental concept, it should not Methodologies for modelling demand be decoupled from economic demand, as it should not Needs (or potential demand) This method determines the ignore the budget constraints of the economy. In fact, the effect of health diseases, epidemiological patterns and country may not have the ability to provide all the health- overall mortality and morbidity rates in the demand for care services presumed to fully satisfy needs. If the area health services and obtains an approximate number of delimited by B (cf. Fig. 1) is larger than the domestic prod- personnel hours required to cover those needs. Needs are uct of the economy, it will be impossible to meet all the usually assessed by a panel of experts in epidemiology and perceived health-care needs of the population. Like any may not match the services that the public wants. other problem involving scarce resources, a serious anal- Economic (or effective demand) In this method, we look at ysis should not abstain from recognizing the existence of the services actually contracted by the population, subject financial impediments. Conversely, it should try to quan- to the usual economic constraints that may put an upper tify needs, serving as a theoretical benchmark for the bound on the quantity solicited. In sharp contrast to the future. first method, effective demand may not imply a healthy This has not always been the case. Some studies esti- population, especially for poor countries without a sub- mate demand solely based on the current level of service sidized health-care service since the general citizen lacks in relation to future projections of demographic profiles the means to obtain health-care services. The method [27, 28], thereby leaving out an important determinant of ignores needs or wants and assumes that all the remain- demand, the epidemiological needs [29, 30]. When and ing variables remain constant, although that requirement how disease trends evolve is critical to properly anticipate may be relaxed by complementing the results with other the needs of the population, a proxy to the expected future methods. demand. For instance, chronic diseases have been increas- Service targets Service targets extend a needs-based ing globally [31]. China, a country usually not associated approach by incorporating other measures, such as con- with overweight and obesity problems, has experienced sumerneeds,inordertoestablishservice-targetratios an upsurge in type two diabetes. According to the data to be accomplished. Service-target approaches decouple reported, in 1980, less than 1 % of Chinese adults had the multiple areas of health-care services and proceed diabetes, but by 2008, the prevalence of the disease had with an independent analysis of each subsystem, with already reached 10 % of the population [32]. As a result, it the main advantage being a more detailed proposition of Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 7 of 19

the changes required, with separate recommendations for the result of initiatives promoted by governments and distinct areas. international organizations to address their own domes- tic shortages of physicians and nurses, while others are ad Evolution of the field hoc contributions of attentive researchers keen on provid- Although the health workforce has long been a concern ing an insightful contribution. The techniques employed to policy makers, including those of ancient Rome [34], vary from descriptive to predictive or merely comparative the first academic research articles discussing manpower techniques and usually involve econometric regressions, planning in general, and health-care workforce planning static tables, linear programming or benchmarking. These in particular, date back to the 1950s. This was a natu- techniques are then applied to the areas of analysis previ- ral response to both the creation of national health-care ously described, either supply, economic demand, needs systems and universal insurance schemes. and service-target or worker-to-population ratios, which A universal health-care system with no exclusion based we will identify next. on preconditions and with no restrictions on access, an A significant part of the research papers produced idea put forward by Bismarck in the compulsory social at that time are well-documented, with comprehensive insurance form, and promoted by Beveridge as a national lists and reviews of the models developed still available health service [35], requires a well-prepared and readily [40, 41]. Of these, we highlight those that are still cited in available team of physicians, nurses and administrative the literature and available online. staff. To ensure that services are in fact provided, public medical universities were created along with subsidized Supply-based methodologies access to medical training. These reforms resulted in the The very initial concern of those conducting HHR plan- emergence of a national ecosystem of health-care suppli- ning was estimating the necessary number (head count) ers and a pool of patients, a significant change from the of medical professionals to either maintain the current decentralized network of health-care providers. The ubiq- worker-to-population ratios or reduce/increase it if an uity of access required providers to be distributed evenly imbalance was found. One of the first insights into the so as to satisfy the needs of the population. evolution of the supply of physicians was done by crossing After this period of sustained and prolific economic the observed physician-to-population ratios along with growth, a period of crisis followed. Expectably, the eco- the posited population growth in the United States of nomic slowdown put the focus on efficiency, towards a America, by that time impulsed by the “baby boom” and better use of the available resources. During this period, by an expected increase in the use of medical services. many developed and developing countries experienced ThepeopleinchargeofHHRplanningevaluatethenum- shortages of health-care providers, mostly nurses [36], ber of physicians required to maintain the ratios given justifying the growing interest in this newborn academic those demographic and economic changes [42, 43]. In research field. the report, the same criterion is used to estimate future This was the period when the first articles on health- manpower requirements for all the available medical spe- care workforce planning emerged. We separate the analy- cialties, nurses and miscellaneous professions necessary sis of the unfolding of HHR planning into three separate for due operation. stages, corresponding to the evolution of how the health- One way of doing so is to look at the current stock of care worker is perceived as an object of study [37]: (a) professionals and factoring in negative and positive flows the health worker as a production factor, (b) the health that affect the stock. Factors such as mortality, migra- worker as an economic factor and (c) the health worker as tion or retirement generate losses to the current work- a necessary resource. This structure is helpful in the sense force stock. Likewise, entries from medical schools and that it exposes the role given to the workforce, once stud- immigration increase the current level of professionals. ied as an inorganic fixed-input factor and more presently Models that map this structure are commonly known as viewed as a complex and necessary resource with its own “stock-and-flow”.Despite not using this specific terminol- idiosyncrasies like any other economic agent. ogy, models created at the time already incorporated the idea of increases and decreases in the current stock due First phase: factor of production to exogenous factors and then used that information to The first articles published on the subject date back to obtain projections [44–46]. 1950, with HHR planning being perceived as a production Focusing particularly on the supply of nurses in the function, where the labour workforce is an input factor. United States of America, other papers proceed with an The research, triggered by general health worker short- analysis of the economic factors, namely the hourly wage ages in developed countries [38, 39], led a growing and and the wage of the nurse’s spouse and the effect on the diversified body of research that diverged into different supply of nursing professionals [45, 47]. Evidence sug- approaches. Not surprisingly, some of these articles are gested that hospitals exercise monopsony power, which Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 8 of 19

has an impact on how a supply gap may be tackled. number of general practitioners, medical specialists, avail- Moreover, results also suggest that the cost of paying able short-term general hospital beds, admissions and wage incentives to increase working hours is considerably mean duration of stay per case [54]. This approach is smaller than the cost of training additional profession- also similar to the one used in two other models, the als, something to take into consideration when evaluating first using data aggregates to facilitate HHR planning at HHR reforms. national, state and substate levels and the second going The product of this novel research was tested in the to the level of detail of the individual and his interactions field. For instance, in the analysis of the health-care work- with professionals and institutions [46]. force in Taiwan, estimates for the supply were generated More comprehensive approaches to estimate economic on the basis of retirement, migration and death rates (effective) demand were also addressed. Some papers sug- applied to graduations. They incorporate the training pro- gested incorporating indicators such as an increase in cess and its effect on the supply of physicians [48]. population, economic development, improved education, Methodologies: Training (entries and losses) [42–46, 48], a change of supply, age distribution and other unpre- Productivity [45, 47], and Worker-to-population ratios dictable factors. Simple calculations, such as the ones used [42, 43]. in the former Soviet Union, could be performed by extrap- olating based on observed norms of practice regarding the Demand-based methodologies number of patients attended and then complemented with One of the first publications in the field of HHR planning basic biological needs by incorporating data about mor- starts by differentiating the aforementioned dimensions bidity and mortality rates [44]. Methods like this were then of workforce planning [49]. Klarman et al. argue that, applied to countries such as Taiwan, characterizing cur- although medical needs could form the basis for determin- rent public and private sector demands for health services ing workforce requirements, it cannot be decoupled from [48]. economic costs, an active constraint to the extent, scope Another option for measuring demand also elaborated and applicability of reformist policies. A forecast of the during this time consisted of using other indirect indica- necessary supply of physicians is not provided, but it is tors, namely short-stay services, services of nervous and suggested that the shortages in the specialty areas may be mental hospitals, physicians’ services outside hospitals, a sign of an overall supply shortage. dental services and other health services. The data is then Another way of predicting the necessary future hospital fed into a model that tries to minimize the gap between beds is by extrapolating from a set of factors assumed to the number of individuals employed in medical services drive the demand for health care, namely socio-economic that attend to the demand for personnel in that occupation factors and biologic needs, measured through morbidity [53]. Estimates were generated for the United States. rates [50]. This approach was also used to estimate hospi- Finally, it should be noted that attention was constantly tal bed requirements, providing both empirical works on being drawn to the importance of prevailing morbidity, a real data for the United States [51] and theoretical frame- basic indicator for assessing medical manpower based on works with hypothetical parameters [52]. In some cases, aneeds-basedapproach.Someauthorsstressthatitisthe the approach of forecasting bed requirements would be hospitals and their internal need for residencies that actu- extended to other health-care units such as primary med- ally determine the number of specialties [55]. This may ical care, nursing home care, consultant medical care not reflect with accuracy the actual needs of the popula- (medical care provided by a physician with specialized tion since patients could potentially remain unattended or training), hospital care or domiciliary care [52]. in long waiting lists, but it is an insightful indicator if wait- Methods for estimating the number of professionals ing lists are also factored in. Finally, they also consider the required (head counts) from a demand perspective also specialty of the physicians’ role, warning that general prac- started emerging at around this time. For instance, in one titioners fulfil key medical functions and should not be case, estimating the number of necessary physicians for relegated to second place. The concept of skill mix,despite the future was done by calculating the number of profes- not formally and explicitly defined, is here put in evidence. sionals necessary to close the gap between observed and Methodologies: Needs (potential demand) [44, 46, 48– unattended demand, where demand is measured in terms 50, 55], Economic (effective demand) [44, 46, 49–54], and of utilization. In this case, using service-level indicators Service targets [46, 53]. again for the United States [53]. In other studies targeting the U.S.’s health system, the Second phase: economic agent influence of exogenous variables such as age, income and The first phase of HHR planning was characterized mainly urbanization is used to extrapolate the effect of dependent by an aggregate analysis of the health-care market, with variables on health policy and HHR planning, includ- independent and/or cross-analysis of supply and demand. ing the number of persons with health insurance, the Reviews produced at that time refer essentially to needs- Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 9 of 19

and demand-based approaches, as well as simple worker- also be used to assess the physicians’ productivity, a study to-population ratio benchmarks [56]. The phase that conducted using data from the United States [67]. starts in the late 1970s and goes onward through the 1980s In the same line of research, some authors conducted an and 1990s redefines the role of the HHR, previously seen observational study of 56 physicians in order to uncover as an homogeneous input factor, into a complex economic the factors that may influence productivity, measured as agent [37]. The adoption of such perspective broadens the the ratio between the number of patients seen per physi- scope of analysis, namely by assuming that health-care cian and the time spent with the patient [24]. The main workers react to economic incentives. research question was understanding which factor con- The deepening of the analysis is done through the appli- tributed the most to the variance in productivity: the cation of microeconomic theory to the study of health patient or the physician. Results suggest, according to the labour workforce, thereby exposing dimensions that had study conducted in a Veteran Affairs’ medical centre in gone unnoticed when looking only at the aggregates, the United States, that the individual physician explains although a macroeconomic analysis continued to take the variations in productivity observed, with the actual place [57]. It was triggered by two macroeconomic obser- patient playing a minor role. Similarly, in another study vations occurring at this time [37]: a perceived oversup- also conducted in the United States, the productivity of ply of physicians and nurses [58–60] and an upsurge in physician assistants and nurse practitioners and their role health-care expenditures [8]. During this phase, atten- in the health-care workforce is analysed [68]. Scheffler tion was given to topics such as health worker licen- et al. find that these two categories of health workers sure [37, 61], information asymmetry distortions [62] and could have a significant influence on the future health- its potential repercussion as an unnecessary increment care workforce if some vertical and horizontal substitu- in demand induced by health suppliers [63] and health tion occurs and tasks are delegated. Note that the change worker performance and productivity [64]. Furthermore, of setup hereby suggested tackles productivity from a dif- HHR planning became a major concern in related fields, ferent angle: instead of raising the output, the inputs are such as dentistry [65]. altered. Methodologies: Productivity [14, 24, 47, 64, 66–68] and Supply-based methodologies Skill mix [68]. Although the previously mentioned topics are of notable relevance, some have no direct utility in the elaboration Demand-based methodologies of projections and forecasts of future health-care needs, Studies focusing solely on the demand side produced dur- serving only for policy guidance. For that reason, we ing this phase are considerably less common than in the will concentrate our efforts on the performance and pro- first phase. The ones that do so are more concerned with ductivity of health workers, a method fully within the the lack of attention given to the importance of biologi- umbrella of supply. In terms of policy, it is less demanding cal needs. It is interesting to note that, at the turn of the to put in practice as it does not require structural changes decade and in subsequent years, a lot of emphasis is again to the training process or to medical schools. In theory, put on the needs of the population. Some authors sug- more people can be served with the exact same amount gest a needs-based evaluation as a requirement to produce of human resources if only their productivity increases. accurate forecasts [29, 56]. This option contrasts with that Improving the efficiency of the available pool of resources of other authors, which propose using benchmark as a is therefore an attractive methodology. viable alternative to potential or effective demand projec- This is the line of research followed in a paper where tions [69]. The work developed consisted of comparing a microanalysis of the factors that may influence the out- thenumberofactivephysiciansper capita in the United put (and therefore productivity) of the health workers is States, adjusted for population differences between simi- conducted, in particular nurses in the United States [47]. lar locations, without uncovering the causes for the given Sloan et al. found that there is a strong supply response asymmetries. to the hourly wage. Raising the hourly wage is, in fact, Assessing the needs of the population was also the their proposal to respond to a short-run supply shortage, method of choice in the dentistry field to calculate oral arguably a quicker response than changing the number of health workforce requirements. In particular, needs were intakes to nursing schools. Taking another route to reach projected by the amount of oral care, including preven- the same goal, one study tries to undercover job satisfac- tive, special group care, surgical, orthodontic, periodon- tion indicators and perceived productivity in 24 hospitals tal, restorative and prosthetic, that different age cohorts for a staff nurse population [66]. The purpose is to under- would require [70]. Then, the time necessary to treat each stand the factors that may raise productivity but also to of these conditions is estimated, and the number of den- find a connection between job satisfaction and the quality tists to perform those tasks is derived. Also applied to of care provided. Similarly, waiting and distance times can dentistry but with a focus on the skill-mix distribution, Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 10 of 19

productivity changes are estimated by examining role sub- and epidemiological) and supply inputs. Furthermore, it is stitution in dentistry [71], helping to conduct evidence- continuously stressed that the epidemiological drivers of based scenario analyses in The Netherlands. the need for health-care services should always be part of Methodologies: Needs [29, 70], Skill mix [71] and HHR planning [30, 74]. Worker-to-population benchmarking [69]. When looking at the research literature produced at the turn of the century, this trend becomes clear. Summing up the results achieved so far, we can see that health- Integrated methodologies care workforce planning is a complex endeavour, and it A new strand of the literature also emerged during this becomes necessary to identify all the relevant variables to phase covering supply while at the same time consider- accurately forecast the necessary resources for the future ing projected changes to demand. In a review of supply [75]. Again, these variables relate to supply and needs projections conducted both in Canada and in the United methodologies. A practical work conducted in Lithua- States [14], the authors argue that the traditional supply nia to forecast family physicians for a 10-year timespan projection methodology that characterizes the licensure employs this approach [76]. Firstly, this approach calcu- cycle and productivity metrics is incomplete if unmet lates the supply of physicians through the usual process of needs of the population are not defined and included as modelling the training of physicians. Moreover, it crosses a clear research goal, as well as economic, financial or the supply forecasts with three different projections for infrastructure resource constraints. demand: firstly, the requirements established by a panel of The integrated approach is also present, for instance, experts using a Delphi technique; secondly, the resources in the implementation of the “System for Health Area necessary to increase the number of visits; and thirdly, Resource Planning” (SHARP) [72]. This analytical frame- an upper bound placed on the worker-to-population ratio work combines all the major methodologies: it includes so that one family physician serves no more than 3 000 the socio-economic factors that drive economic demand, inhabitants. The conclusions reached suggest that the morbidity and the remaining epidemiological factors that well-informed panel of experts elaborated the most accu- drive needs, the formation process of the health-care rate projection of demand for family practitioners and that supply of workforce and utilization rates in order to incor- none of the supply projections was right on target. Simi- porate the current use of health-care services. The frame- larly, in a forecast analogous to the nursing profession in work was successfully used to support HHR planning in Germany, the analysis is extended from the usual supply Canada, especially in the province of Ontario, reinforcing and demand to include the effects of occupational flexibil- the idea that an integrated or systems approach, combin- ity and employment structure. Adding these two elements ing the multiple facets of the problem, is the way to go in to the analysis has a relevant influence on the projections the future. [77]. Notably, this pensiveness with the organizational Methodologies: Integrated [14, 72]. role, where the HHR is more than an aggregate number but rather a dynamic and complex sum of individuals, is clearly gaining traction. Third phase: fundamental resource In the same line, some researchers suggest a needs- In this phase, the notion of health labour workforce is based analytical framework that incorporates input from reformulated, this time viewing it as a necessary resource. four separate elements: demography, epidemiology, stan- From the 1990s onto the 2000s, the emphasis is on the dards of care and provider productivity [30], again falling regional asymmetries in the placement of the workforce in the realm of integrated approaches. Alternatively, needs and in the migration flows from developing to developed can be decoupled in a functional form so that service countries [37]. All models proposed include both supply- targets can be defined and deployed [1]. Dreesch et al. and demand-based methodologies to tackle the problem. claim that methods focusing strictly on the supply, on the demand or on both fail to address or recognize the effects of the skill mix (the potential of substitution) between Integrated methodologies health professions. The importance of a more integrated Methodology-wise, the trend observed is a continuation approach to HHR planning is also restated. With more or of the second phase, with the call for a holistic approach less variables, the trend is clear: recent models use infor- to the problem. HHR planning must be addressed from mation from both demand- and supply-based method- an integrated perspective, including when analysing all ologies, including inputs as varied as demography, the the blocks of the functioning system so as to calcu- training process, workers’ productivity or biological needs late the current and future gap between supply and in order to generate their forecasts [18, 78, 79]. demand [73]. The authors’ proposal is in line with the Although the emphasis is fundamentally put on address- SHARP framework: modelling key demand (economic ing the problem from an integrated perspective, new Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 11 of 19

strands of literature were also developed during this Discussion phase. For instance, it is suggested that instead of Five decades of work in HHR planning fuelled by eminent addressing the problem from a quantitative perspective, global shortages of health professionals have contributed either by adding to or subtracting from the stock of health to establishing this research field as an important scien- workers, it should rather be addressed with internal reor- tific area, decisive for achieving worldwide health-care ganizations, redefining which tasks can be performed by targets [1]. Significant results have been attained. In par- whom [80]. Such internal substitution and activity dele- ticular, new methods and techniques were developed, and gation could be executed by transferring skills from the the accuracy of projections improved remarkably [23], and medical specialist and the general medical practitioner HHR planning became an area of prominent interest, with to other health professional roles, namely nurses with the number of publications in the field increasing over higher education (midwives) or by creating new roles. the years. Moreover, the literature evolved, replacing some This methodology involves, therefore, playing with the approaches with others, paying more attention to the skill mix of the health-care professionals. This was put in health-care workers and their productivity and to the del- practice in Ireland by employing a model that targets both egation and distribution of skills. It prioritized integrated supply and demand, reflecting the concerns for including approaches and the role of epidemiology in addressing the all parts of the system [28, 81]. Moreover, it tests four pol- problem. In fact, when we look through all the methodolo- icy interventions, three of which related to supply and the gies reviewed (Fig. 2), the emerging trend clearly supports last related to the skill mix: increasing vocational training this claim. Integrated approaches are gaining ground after places, recruiting professionals from abroad, incentivizing decades of partial analyses turning to either a supply- or later retirement and increasing nurse substitution so that a demand-based approach and in its simplest form only nurses can deliver more services. Similar studies, encom- resorting to worker-to-population ratio benchmarks. passing the workforce supply, demand and the skill mix, In Table 4, we summarize the methodologies and were also conducted in the dentistry field during this describe the necessary assumptions for using each of phase [82]. In this case, workforce supply and demand for the approaches, along with their advantages, limitations, oral health needs are projected to study the impact of skill- how these limitations are overcome, requirements and the mix reorganizations. To forecast future dentist numbers, countries in which their usage was documented (accord- a simple percentage increase based on previous yearly ing to [9]). In the past, this overview would probably help increases is considered. To estimate demand, demogra- in choosing the methodology to adopt. With the call for phy evolution, rates of edentulousness, patterns of dental more integration, it assists in showing how a methodology attendance and treatment rates of older people, as well may fill in the gap towards a cohesive framework. Also, as general dental service treatment times, are considered. it serves to show that there is no perfect methodology The effect of the skill mix is then studied considering sev- capable of providing accurate forecasts without consider- eral scenarios of varying skill-mix use. Gallagher et al. find able pitfalls and that there is a trade-off between simplicity that widening the skill mix can be extremely helpful to and completeness, where going for a simpler methodology build capacity for dental care. may implicate leaving out important parts of the problem. Another concern that is raised during this phase is that of measuring the outcome as an important indicator for An integrated approach assessing the quality of the health-care services. The out- The importance of a comprehensive, integrated approach come is a fundamental indicator for HHR planning. In is continuously emphasized throughout the period in particular, equitable and timely access to health care are a review [3]. Although the need for an integrated approach precondition to a good outcome, which is the variable to had already been stressed in several past publications, it be maximized [83]. keeps on reappearing, suggesting that it might not have In summary, it can be said that this stage was a phase been fully addressed as of yet. This approach faces many of settling with methodologies, namely supply-, demand- challenges. A dynamic, system-level perspective covering and needs-based approaches, and of urging for a more key drivers of supply and demand that includes both man- integrated approach while paying attention to the roles power planning and workforce development is critical to of each health professional and the degree of substitution overcome such challenges [81]. The importance of paying between professions. Furthermore, a concern about the attention to needs is also continuously stressed, as changes outcome of health-care services was raised, where effec- in the health patterns of the populations take place [84]. tiveness and quality of the treatment is considered on par In summary, integrated approach refers to a method that with the number of patients seen (productivity). incorporates in its process projections of the workforce Methodologies: Integrated [18, 18, 28, 30, 73–79, 81, 82], supply and the impact of microeconomic and organiza- Skill mix [1, 28, 77–82], Needs [30, 77], Service targets [1] tional changes in productivity and in the skill mix, of the and Productivity [77–79] evolution of demand for health-care services and also of Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 12 of 19

Fig. 2 Identification of the conceptual methodologies found in some of the literature for the period of 1950–2013 the evolution of health diseases and its potential impact process so as to obtain an initial snapshot of the cur- on the health system. rent workforce. The current stock, which may or may not Notwithstanding, integrating all the pieces may be a be enough to tackle current demand, in which case an puzzling task. To assist with the task, in Fig. 3, we pro- imbalance exists, is subject to positive and negative flows vide a high-level functional diagram with a proposal for that may alter its number and composition. This given how methodologies could be coupled so as to turn it into quantity of workers may provide more or less health-care a seamlessly integrated system. On the supply side, we services depending on their productivity and skill mix, have the current stock of workers along with the training and that influences the conversion from head counts to Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 13 of 19 a usage Australia, Belgium, Canada, Chile, Denmark, Finland, France, Germany, Ireland, Israel, Japan, South Korea, Norway, Switzerland, The Netherlands, United Kingdom, USA Switzerland, United Kingdom, USA Australia, Canada, Japan, Korea, Netherlands, Norway, Netherlands, United Kingdom Chile, France, Ireland, Israel, Switzerland, United Kingdom Requirements Documented Accurate and up-to- date accounting of the current stock of physicians and nurses, migration rates, entry and drop out rates and expected retirees Service usage levels from the health-care sector given number of FTEs (or head counts) Operational indicators like the number of patients served with a Education schools that can provide advanced education to the existing workforce Records of the current workforce to popula- tion ratios limitations Incorporate a model of demand: economic or needs-based (or both) Evaluate current level of service through waiting lists, overtime hours, foreign workers, etc. Do not precludeevaluating the from number of professionals necessary given different productivity levels Providing success stories to involved stakeholders, health authorities and medical associations Does not take into account the intrinsic differences between regions and countries, the productivity and skill mix of the available workforce Demand for medical services is assumed to remain constant, which may not be true No critical assessment of the adequacy of current service levels Productivity improvements may not be enough to accommodate large gaps in the supply of professionals Enforcing such changes can be a political challange. Does not solve large gaps in the supply Does not take into account the intrinsic differences between andregions countries, the productivity and skill mix of the available workforce Predictions for the future supply can be obtained in a fairly simple and immediate way Does not require a change in the quantity of human resources. Can be implemented immediately Does not require a change in the quantity of human resources. Can be implemented immediately Extremely easy to understand and apply Useful for providing baseline comparisons supply gap Demand for medical services is assumed to remain constant and the projections are used to reduce the Physicians and nurses act as rational agents and react to economic incentives like wage increases Professionals can assume new roles and perform new tasks Regions and/or countries can be directly compared dropouts), migration flows, attritions andrates retirement health-care professionals based on the current stock of clinicians, the training process (entries and economic incentives to promote higher productivity. Work harder or work smarter vertical (between physicians and nurses) other health professionals. Substitution can be horizontal (between medical professions) or Specifies desirable worker- to-population ratios based on direct comparison with another region of country The methodological approaches established during the first phase of research Training Projects the availability of Productivity Reorganize services and/or Skill mix Delegate certain tasks to Worker-to- population ratios Supply Table 4 Methodology Description Assumptions Advantages Limitations Overcoming Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 14 of 19 Australia, Belgium, Canada, Denmark, Finland, Germany, Japan, Norway, South Korea, Switzerland, The Netherlands, USA Belgium, Canada, Germany, United Kingdom Accurate and long- term demographic estimates Service-usage levels from the health-care sector Macroeconomic indicators and datastatistical crossing income and usage Demographic estimates that are accurate Service-usage levels from the health-care sector Current level of service Evaluate current level of service through waiting lists, overtime hours, foreign workers, etc. Include a needs-based evaluation Take financial constraints into consideration Consider an upper bound for a practical result Consider projections of the most common health patterns Incorporate economic considerations in the model Incorporate economic considerations in the model demand that exceed practical limits No critical assessment of the adequacy of current service levels Ignores the real demand, focusing instead on the effective demand Tends to produce estimates of HHR Absence of economic/efficiency considerations may therender projections unattainable Dependent on epidemiological projections which may not be obvious Does not consider the current level of provision nor the capacity of the country to deliver health care May originate unrealistic assumptions Ignores financial and other active constraints (Continued) to apply Allows decoupling of the various components of demand and their influence on the overall aggregate demand Conceptually easy to understand and Allows for a fine-grained analysis of the requirements of each medical specialty Is independent of the current service- utilization ratios Easy to understand Easy to define, interpretand understand Facilitates cost estimation Requires modest data and planning capabilities Current level of service is adequate. Skill mix and distribution of health service is appropriate Demographic profile of the population and its effect on health- care demand can be accurately forecasted All health-care needs can and should be met Resources are used in accordance to needs Assumes that established service targets are achievable in terms of financial and physical capital resources requirements by projecting the effect of demographic and socio-economic factors on the current level of service epidemiology on the demand for health-care services Projects age- and gender- specific needs based on morbidity epidemiological trends for the production of health-care services, which are then converted to HHR requirements The methodological approaches established during the first phase of research Economic Estimates future Needs Considers the effect of Service targets Defines normative targets OECD Report Source: adapted from Hall and Mejia [13], O’Brien-Pallas [11] and Dreesch [1] a Table 4 Demand Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 15 of 19

Fig. 3 An integrated system that incorporates several methodologies to address the many facets of HHR planning full-time equivalents (FTEs). Such conversion is critical to the availability of the data increases the probability of a properly assess the health-care workforce, as a significant more comprehensive projection. number of physicians and nurses work part-time only. For Simpler approaches require fewer data. Worker-to- this reason, FTE is a more accurate measure as it nor- population ratio benchmarks require a head count of malizes the head counts. On the demand side, economic the number of licensed medical professionals, usually (effective) demand can be initially measured by analysing made available by the government, medical and nurse utilization indicators. How this demand will evolve in the associations or by unions. Service targets use the cur- future will then be subject to typical economic factors rent level of service, which can be obtained from the such as demography and the growth of the income/GDP. hospitals’ operational key performance indicators. Needs In parallel, potential needs can be assessed by incorporat- (potential) and economic (effective) demand, on the other ing incidence and prevalence of diseases and then map- hand, require a more extensive set of indicators. For ping a given disease to an estimate of FTE requirements. needs, it is necessary to assess and validate current and Whether future supply forecasts should tackle all of the future incidence and prevalence of diseases and how that estimated needs is a decision left to the consideration of may convert into necessary resources. Both tasks are the policy maker, as this analysis does not incorporate not straightforward and usually require acclaimed experts financial constraints. Such an integrated approach is more in epidemiology to step in and provide both the esti- complex, but not necessarily more difficult [12]. In fact, mates, as well as an accounting of the resources that policy-making cannot abstain from factoring in financial will be necessary. Effective demand makes it necessary and service planning considerations in a post hoc analysis, not only to obtain metrics similar to those indispens- since there may not be enough resources to accommo- able for a service-target analysis (such as the number of date for a sudden increase in the number of professionals. inpatients and outpatients, number of occupied hospital Such analysis is not limited to a money perspective, to beds, average length of stay) but also demography and the financial burden inputted on the system for educating socio-economic projections and how they affect demand. and hiring these medical professionals or to the installed Finally, modelling supply is also a challenging task in capacity in terms of medical schools, university hospitals, terms of data requirements. Unless evidence is found hospital beds, primary care facilities and others, in order showing that the worker-to-population ratios will remain to absorb planned increases in the health-care services constant for a long period of time, a supply-based anal- labour market. ysis must be factored in. In such a case, it is necessary to know the current stock of licensed providers, as well Data requirements as the number of intakes, exits and annual attritions, None of these methodologies can be applied without the which makes it necessary to model the training of medical adequate data to feed the model. A bare minimum of professionals. information regarding the available medical workforce is Assuming that developing countries are in possession of always required. Table 5 summarizes the most impor- fewer data and that developed countries have more infor- tant indicators for conducting a proper forecast. It is not mation available, methodologies that require an extensive strictly necessary to possess all the information listed, but set of data will be difficult to implement in developing Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 16 of 19

Table 5 Data requirements for making use of each of the different documented methodologies Methodology Indicators Data requirements Stock of licensed providers Baseline stock, age/sex distribution, growth projections Annual additions to licensed stocks Graduates, in-migration (foreign-trained, immigrantes, on temporary work permits), returned to profession Supply High Education/training programmes Number of programmes and students enrolled, attrition rates, years to complete programme, number of graduates, costs Annual attritions to licensed stocks Retirements, mortality, career changes, emigration, abroad Labour market Occupational participation rates, occupational employment rates, employment Productivity projections, vacancy rates, turnover rates, wage rates, productivity growth, cyclical factors, High alternative career options Employment status Full-time, part-time, casual, full-time equivalent (FTE), average hours worked, direct patient care hours, no longer practising, not licensed in jurisdiction Government policy variables Skill mix HHR education funding, alternative delivery modes, licencing regulations, professional High roles/deployment, recruitment/retention strategies, immigration policy, remuneration rates/types, HHR capacity-building Health labour workforce Worker-to-population ratios Number of active and employed physicians and nurses Low Population demographics Total population, age/sex distribution, births/deaths, population projections Economic Socio-economic variables High Disposable income, GDP growth projections, ethnic factors Population health status Age/sex mortality, morbidity, acuity Needs Epidemiology High Incidence and prevalence rates, hospital discharges, health patterns of the population Utilization patterns Service targets Number of occupied beds, number of inpatients and outpatients, number of Low to high surgeries/screenings/consultations performed, etc. countries. Therefore, such countries may start by using the worldwide supply of medical professionals, an assess- simple techniques such as the worker-to-population ratio ment of what has been done and achieved and what or service-based benchmarks to tackle their present remains to be done was necessary to properly guide fur- imbalances. Developed countries should continue collect- ther developments in this relevant field. Moreover, when ing data and enhancing their models, adding less tangible we contemplate the complex training process required and yet relevant dimensions, such as productivity or skill to earn a licence as a practitioner, we understand that mix if they are not present already. a shortage in medical professionals cannot be accom- modated fast enough by decree, either by increasing the Conclusion number of intakes to medical schools or by inviting more In this paper, we reviewed over 60 years of publications foreign-trained doctors or nurses. in HHR planning. While doing so, we observed the evolu- Despite the abundance in approaches and techniques to tion of the field, when and how methodologies emerged, determine supply and need for professionals, none of the how they have been applied and the robustness of the methodologies has ultimately proved to be superior [85]. results, and we also identified the current trends in the Recent studies testing current forecasting models show field. This work was called for because there is still no that there is still plenty of room for improvement given accepted methodology to address HHR planning. Given the gap between projected and actual results [12]. the rampant costs in the health-care sector and the over- It becomes even clearer that workforce planning should all influence that health care has on the general welfare be accurate and performed in due time given the attri- of society, as well as the potential impact of shortages on tions and the delays in enacting policies in the health-care Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 17 of 19

sector. Adapting medical schools, altering legislation and References changing roles is an effort that may take years to bring 1. Dreesch N. An approach to estimating human resource requirements to achieve the Millennium Development Goals. Health Policy and Planning. forth. Therefore, planning has to target a long enough 2005;20(5):267–76. time horizon if it is to be useful and applicable and has to 2. Ozcan S, Taranto Y, Hornby P. Shaping the health future in Turkey: a new be done pre-emptively. role for human resource planning. The International Journal of Health Planning and Management. 1995;10(4):305–19. It now seems obvious that, like any other complex prob- 3. Roberfroid D, Leonard C, Stordeur S. Physician supply forecast: better lem, all the determining pieces of the system and their than peering in a crystal ball? Human Resources for Health. 2009;7(1):10. interdependent relationships must be duly accounted for. 4. Birch S. A re-examination of the meaning and importance of Therefore, pressing for integrated approaches is still a supplier-induced demand. J Health Econ. 19931–22. 5. WHO. The World Health Report 2006: working together for health. valid and up-to-date concern. Furthermore, envision- Geneva: WHO; 2006. ing the health worker in its entire complexity makes 6. Williamson AM, Feyer AM. Moderate sleep deprivation produces it possible to address the problem more comprehen- impairments in cognitive and motor performance equivalent to legally prescribed levels of alcohol intoxication. Occup Environ Med. 2000;57(10): sively, leaving room to improvements in productiv- 649–55. ity and in the distribution of work without having to 7. Steinbrook R. Private health care in Canada. N Engl J Med. 2014;354(16): directly interfere with the training process or with the 1661–4. 8. Huber M. Health expenditure trends in OECD countries, 1970-1997. health providers. Operations research and lean manage- Health Care Financ Rev. 1999;21(2):99–117. ment are particularly relevant in this area. This strat- 9. Health at a Glance 2013 OECD Indicators: OECD Indicators, 2013edn. egy may be, in fact, a first attempt to solve the lack of Health at a Glance 2013 OECD Indicators. OECD Publishing; 2013. professionals. 10. Ono T, Schoenstein M, Lafortune G. Health workforce planning in OECD countries. Technical report OECD, France. 2013;62:131. The results of our review point in one clear direc- 11. O’Brien-Pallas L, Baumann A, Donner G, Murphy GT, Lochhaas-Gerlach J, tion: accurate HHR planning requires an approach that Luba M. Forecasting models for human resources in health care. Health is both integrated and flexible, featuring supply and and Nursing Policy Issues, 1–10. 2001 21. 12. Van Greuningen M, Batenburg RS, Van der Velden LF. The accuracy of demand (potential and effective) and incorporating less general practitioner workforce projections. Human Resources Health. tangible factors, such as skill mix and productivity. 2013;11(1):1. The road to accurate HHR planning cannot abstain 13. Hall T, Mejia A. Health manpower planning: principles, methods, issues. from this. 19781–146. 14. Lomas J, Stoddart GL, Barer ML. Supply projections as planning: a critical review of forecasting net physician requirements in Canada. Soc Sci Med. Endnote 1985;20(4):411–24. aHenceforth, the term ’approach’ is used loosely to refer 15. Dussault G, Buchan J, Sermeus W, Padaiga Z. Assessing future health to the conceptual methodology employed rather than to workforce needs. 20101–46. http://www.euro.who.int/__data/assets/ pdf_file/0019/124417/e94295.pdf. the technical and scientific apparatus used to obtain a 16. Dall T. The physician workforce: projections and research into current projection or forecast. issues affecting supply and demand. 20081–111. http://bhpr.hrsa.gov/ healthworkforce/reports/physwfissues.pdf. 17. Chopra M, Munro S, Lavis JN, Vist G, Bennett S. Effects of policy options Abbreviations for human resources for health: an analysis of systematic reviews. The HHR: Health-care human resources; OECD: Organisation for Economic Lancet. 2008;371(9613):668–74. Co-operation and Development; WHO: World Health Organization. 18. Scheffler R. Forecasting the global shortage of physicians: an economic- and needs-based approach. Bull World Health Organ. 2008;86(7):516–23. Competing interests 19. Schofield DJ, Fletcher SL, Callander EJ. Ageing medical workforce in The authors declare that they have no competing interests. Australia - where will the medical educators come from? Hum Resources Health. 2009;7(1):82. 20. Folland S, Folland S, Goodman AC, Goodman AC, Stano M, Stano M. Authors’ contributions The economics of health and health care, 2007th ed. Pearson Prentice MAL proceeded with the literature review and drafted the paper, with BAL and Hall, USA; 2007. ASA providing guidance, critical assessment and peer review of the writing. The three authors read, reviewed and approved the final manuscript. 21. Lakhan S, Laird C. Addressing the primary care physician shortage in an evolving medical workforce. Int Arch Med. 2009;2(1):14. 22. Maynard A. Medical workforce planning: some forecasting challenges. Acknowledgements Aust Econ Rev. 2006;39(3):323–9. We are extremely grateful to all the reviewers for their insightful comments 23. Van Greuningen M, Batenburg RS, Van der Velden LF. Ten years of health and contributions, as they significantly contributed to the improvement of this workforce planning in the Netherlands: a tentative evaluation of GP paper. Also, we would like to thank several members of the European planning as an example. Hum Resources Health. 2012;10(1):1. Operations Research Society and the scientific committee of the EURO 24. Smith D, Martin D, Langefeld C, Miller M, Freedman J. Primary care Operational Research applied to Health, which provided insightful ideas and physician productivity: the physician factor. J General Internal Med. feedback on the ongoing work. 1995;10(9):495–503. 25. Grossman M. On the concept of health capital and the demand for Author details health. J Pol Econ. 1972;80(2):223–55. 1 INESC TEC, Faculdade de Engenharia, Universidade do Porto, Porto, Portugal. 26. Culyer AJ, Wagstaff A. Equity and equality in health and health care. J 2 Faculdade de Economia, Universidade do Porto, Porto, Portugal. Health Econ. 1993;12(4):431–57. 27. Al-Jarallah K, Moussa M, Al-Khanfar KF. The physician workforce in Kuwait Received: 10 November 2014 Accepted: 2 May 2015 to the year 2020. The International Journal of Health Planning and Management. 2010;25(1):49–62. Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 18 of 19

28. Teljeur C, Thomas S, O’Kelly FD, O’Dowd T. General practitioner 60. Iglehart JK. From physician shortage to patient shortage: the uncertain workforce planning: assessment of four policy directions. BMC Health future of medical practice. Health Affairs. 1986;5(3):142–51. Serv Res. 2010;10(1):148. 61. Gaumer GL, National Center for Health Services Research. Regulating 29. Birch S, Eyles J. Needs-based planning of health care: a critical appraisal of health professionals: a review of the empirical literature, 1984edn. USA: the literature. CHEPA Working Paper Series. 1991;91(5):. U.S. Department of Health and Human Services, Public Health Service, 30. Birch S, Kephart G, Tomblin-Murphy G, O’Brien-Pallas L, Alder R, Office of the Assistant Secretary for Health; 1984. MacKenzie A. Human resources planning and the production of health: a 62. Blomqvist Å. The doctor as double agent: information asymmetry, health needs-based analytical framework. Can Public Policy. 2007;33(s1):1–16. insurance, and medical care. J Health Econ. 1991;10(4):411–32. 31. Nugent R. Chronic diseases in developing countries. Ann NY Acad Sci. 63. Evans R. Supplier-induced demand : some empirical evidence and 2008;1136(1):70–9. implications. Economics of Health and Medical Care. 1974162–73. 32. Hu FB. Globalization of diabetes: the role of diet, lifestyle, and genes. 64. Reinhardt UE. Physician productivity and the demand for health Diabetes Care. 2011;34(6):1249–57. manpower: an economic analysis, vol. 1, 1975 edn. USA: Ballinger Pub. 33. Murphy GT, O’Brien-Pallas L. How do health human resources policies and Co.; 1975. practices inhibit change? A plan for the future. 20021–36. http://qspace. 65. DeFriese GH, Barker BD. Assessing dental manpower requirements: library.queensu.ca/bitstream/1974/6884/11/discussion_paper_30_e.pdf. alternative approaches for state and local planning. Issues in dental 34. Rosen G. A history of public health. A Johns Hopkins paperback. USA: JHU health policy. USA: Ballinger Pub. Co. 1982. Press; 1959. 66. Kramer M, Hafner LP. Shared values: impact on staff nurse job satisfaction 35. Burau V, Blank RH. Comparing health policy: an assessment of typologies and perceived productivity. Nursing Res. 1989;38(3):172–7. of health systems. J Comp Policy Anal Res Prac. 2006;8(1):63–76. 67. Sloan FA. Access to medical care and the local supply of physicians. 36. Aurousseau P. Distribution of case in Europe - severe shortage of nurses. Medical Care. 1977;15(4):338–46. Can Hosp. 1971;48(10):91–2. 68. Scheffler RM, Waitzman NJ, Hillman JM. The productivity of physician 37. Barnighausen T, Bloom DE. Changing research perspectives on the assistants and nurse practitioners and health work force policy in the era global health workforce. NBER Working Papers. 20091–84. of managed health care. J Allied Health. 1996;25(3):207–17. 38. Guerra M. The shortage of physicians. Jornal do medico. 1965;57(173):643. 69. Goodman DC, Fisher ES, Bubolz TA, Mohr JE, Poage JF, Wennberg JE. 39. Hale T. Why the nursing shortage persists. N Engl J Med. 1964;270: Benchmarking the US physician workforce: an alternative to needs-based 1092–1097. or demand-based planning. JAMA: J Am Med Assoc. 1996;276(22):1811–7. 40. Vector Research Inc, Bonder S, Development U. S. B. o. H. R. Health 70. Bronkhorst EM, Truin GJ, Batchelor P, Sheiham A. Health through oral manpower models: an analysis of health manpower models. DHEW. health; guidelines for planning and monitoring for oral health care: a 1974;1:1–220. critical comment on the WHO model. J Public Health Dent. 1991;51(4): 41. Doyle TC. An inventory of health manpower models volume II. DHEW. 223–7. 1975;2:1–313. 71. Brenninkmeijer OP, Kuitenbrouwer RKJM, Nelissen AMH, Burgersdijk 42. US SG Consultant Group, Bane F. Physicians for a growing America: RCW, Netherlands, Ministerie van Welzijn VeC1, et al. Scenario report, 1959th ed. USA: Public Health Service, U. S. Dept. of Health, Committee on Dental Health Care: future scenarios on dental health care Education, and Welfare; 1959. : a reconnaissance of the period 1990-2020 : scenario report. Norwell, 43. Statistics U. S. B. o. L. Health manpower, 1966-75: a study of requirements Mass.: Kluwer Academic Publishers; 1993. and supply, 1967edn. Report. USA: U.S. Bureau of Labor Statistics; 1967. 72. Denton F, Gafni A, Spencer B. The SHARP way to plan health care services: 44. Baker TD. Dynamics of health manpower planning. Medical Care. a description of the system and some illustrative applications in nursing 1966;4(4):205–11. human resource planning. Socio-Economic Planning Sci. 1995;29:1–13. 45. Altman SH. Present and future supply of registered nurses. MD: US DHEW, 73. Joyce CM, McNeil JJ, Stoelwinder JU. Time for a new approach to 1–168 (1971). J Hum Resources. 1975;10(3):403–406. medical workforce planning. Med J Aust. 2004;180(7):343–6. 46. Yett DE, Drabek L, Intriligator MD, Kimbell LJ. Health manpower 74. Birch S. Health human resource planning for the new millennium: inputs planning: an econometric approach. Health Serv Res. 1972;7(2):134–47. in the production of health, illness, and recovery in populations. Can J 47. Sloan FA, Richupan S. Short-run supply responses of professional nurses: Nurs Res. 2002;33(4):109–14. a microanalysis. J Hum Resources. 1975;10(2):241–57. 75. Ros K, Drzymala L. Planning physician services: is there a method to the 48. Baker TD, Perlman M. Health manpower in a developing economy: madness? Econ Health Policy. 2002;79:1–6. Taiwan, a case study in planning, 1967th ed. USA: Johns Hopkins Press; 76. Starkiene L, Smigelskas K, Padaiga Z, Reamy J. The future prospects of 1967. Lithuanian family physicians: a 10-year forecasting study. BMC Family 49. Klarman HE. Requirements for physicians. Am Econ Rev. 1951;41(2): Prac. 2005;6(1):41. 633–45. 77. Maier T, Afentakis A. Forecasting supply and demand in nursing professions: impacts of occupational flexibility and employment structure 50. Beenhakker HL. Multiple correlation-a technique for prediction of future in Germany. Hum Resour Health. 2013;11(1):1. hospital bed needs. Oper Res. 1963;11(5):824–39. 78. Tomblin Murphy G, MacKenzie A, Alder R, Birch S, Kephart G, 51. Rosenthal GD. The demand for general hospital facilities, 1964edn. O’Brien-Pallas L. An applied simulation model for estimating the supply of Hospital monograph series, no. 14. USA: American Hospital Assoc; 1964. and requirements for registered nurses based on population health 52. Navarro V. A systems approach to health planning. Health Serv Res. needs. Policy, Politics, & Nursing Pract. 2010;10(4):240–51. 1969;4(2):96–111. 79. Masnick K, McDonnell G. A model linking clinical workforce skill mix 53. Maki DR. A forecasting model of manpower requirements in the health planning to health and health care dynamics. Hum Resour Health. occupations, 1967edn. USA: Industrial Relations Center; 1967. 2010;8(1):11. 54. Feldstein MS. An aggregate planning model of the health care sector. 80. Duckett S. Health workforce design for the 21st century. Australian Health Medical Care. 1967;5(6):369–81. Review. 2005;29(2):210–10. 55. Health CP. Estimating need for physicians. Bull NY Acad Med. 1968;44(8): 81. Stordeur S, Leonard C. Challenges in physician supply planning: the case 1068–1084. of Belgium. Hum Resour Health. 2010;8(1):28. 56. Pathman DE. Estimating rural health professional requirements: an 82. Gallagher JE, Kleinman ER, Harper PR. delling workforce skill-mix: how assessment of current methodologies*. J Rural Health. 1991;7(4):327–46. can dental professionals meet the needs and demands of older people in 57. Lipscomb J, Kilpatrick KE, Lee KL, Pieper KS. Determining VA physician England? Br Dental J. 2010;208(3):6–6. requirements through empirically based models. Health Services Res. 83. Astrid Guttmann ECCM. Outcomes-based health human resource 1995;29(6):697–717. planning for maternal, child and youth health care in Canada: a new 58. Grayson MA. Medical educators told to reduce enrollments,. Hospital Med horizon for the 21st century. Paediatr Child Health. 2009;14(5):310. Staff. 1978;7(12):37–40. 84. Tomblin Murphy G, Kephart G, Lethbridge L, O’Brien-Pallas L, Birch S. 59. Schroeder SA. Western European responses to physician oversupply: Planning for what? Challenging the assumptions of health human lessons for the United States. JAMA: J Am Med Assoc. 1984;252(3):373–84. resources planning. Health Policy. 2009;92(2–3):225–33. Amorim Lopes et al. Human Resources for Health (2015) 13:38 Page 19 of 19

85. Ricketts TC. The health care workforce: will it be ready as the boomers age? A review of how we can know (or not know) the answer. Ann Rev Public Health. 2011;32(1):417–30. 86. WHO. Increasing access to health workers in remote and rural areas through improved retention: global policy recommendations: World Health Organization; 2010. 72 p.

Submit your next manuscript to BioMed Central and take full advantage of:

• Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit